Many algorithms have been proposed for multidimensional frequency estimation from a single snapshot or multiple snapshots of data mixture. Most of these algorithms fail when one or more identical frequencies are found in certain dimensions. In this paper, a multidimensional frequency estimation technique from a single datum snapshot is proposed. It applies LU decomposition (Gaussian Elimination) on an eigenvector-based algorithm for multidimensional frequency estimation. This proposed technique is simulated using a MATLAB code. The average root mean square error (RMSE) is investigated as a performance measure of the proposed technique. A comparison between original eigenvector-based (traditional) and the proposed techniques is introduced. The simulation results show that the RMSE of the proposed technique is less than the original one, and it has a more efficient solution for an identical frequency case but at the expense of complexity.

中文翻译：

---

使用基于LU分解特征向量的算法进行多维频率估计

已经提出了许多用于从数据混合物的单个快照或多个快照进行多维频率估计的算法。当在特定维度上找到一个或多个相同频率时，这些算法中的大多数都会失败。本文提出了一种基于单数据快照的多维频率估计技术。它将LU分解（高斯消去）应用于基于特征向量的多维频率估计算法。这项提议的技术使用MATLAB代码进行了仿真。研究了平均均方根误差（RMSE）作为所提出技术的性能指标。介绍了原始的基于特征向量的（传统）与所提出的技术之间的比较。仿真结果表明，所提出技术的RMSE小于原始技术，